Unsupervised Graph-Based Patterns Extraction for Emotion Classification

Survey on Detection of Metal Illnesses by Analysing Twitter Data

“…They use various machine learning algorithms like Maximum Entropy, Naïve Bayes and Support Vector Machines, and check the performance of each algorithm in classifying the tweets into positive negative and neutral emotions. [1] Argueta et al prepared the segmented and normalized data (pre-processed data) to extract patterns of emotion. This was done using a graph based approach.…”

“…[4] approach was to first train the dataset and then test it on a random collection of tweets which showed high accuracy. Argueta et al [1] attempted to classify the data by segmenting it, using the segmented data to create a graph, reducing the graph and finally extracting emotions from the graph. Choudhury et al [3] mainly wanted to look at the moods of the subjects, the degree of physical intensity in an emotion (activation) and the degree of control in an emotion (dominance).…”

“…Pattern of life features revealed the emotional patterns and behavioural tendency of users by finding out the polarity, emotions and interactions they make online. The authors calculated emotional scores of each tweet by using the classifier suggested by [1] Argueta et al, and transformed their 8 emotions into 8 emotional scores. They also predicted the age and gender of the person.…”

Detection of the Prodromal Phase of Bipolar Disorder from Psychological and Phonological Aspects in Social Media

“…We further partition those three categories into five different sub-features: positive ratio, negative ratio, positive combo, negative combo, and flips ratio. • Emotional Scores: Beyond the sentiments, an emotion detection tool proposed by Argueta et al [1] is employed to classify the tweets into eight emotion categories: joy, surprise, anticipation, trust, sadness, disgust, anger, and fear. The emotion classification results are further transformed into emotion scores as follows:…”

“…In general, all of the ensemble features performed well in the onset detection task. The ensemble of Phonological features and BPLF, which is the integration of our proposed features with BD-customized features from original Pattern of Life(PLF), performed the Features(#DIM) 2 mths 3 mths 6 mths 9 mths 12 mths AG (2) 0.475 0.503 0.445 0.434 0.383 Pol (5) 0.911 0.893 0.843 0.836 0.803 Emot (8) 0.893 0.895 0.908 0.917 0.896 Soc (4) 0.941 0.913 0.845 0.834 0.786 LT (1) 0.645 0.589 0.554 0.504 0.513 TRD (1) 0.570 0.638 0.626 0.615 0.654 Phon (8) 0.889 0.880 0.802 0.838 0.821 As can be seen in Table 3, the models trained on three months of user data performed the best, which also indicates that BD features are more obvious when the time period is 2 to 3 months before diagnosis. Rapid cycling bipolar disorder, from DSM-IV, is defined as a pattern of presentation accompanied by 4 or more mood episodes in a 12-month period, with a typical course of mania or hypomania followed by depression or vice versa.…”